Laytus - parlay prediction markets

Laytus is a structured prediction market protocol that solves a fundamental pricing flaw in multi-outcome positions, traditional parlays multiply probabilities assuming full independence, which systematically misprices correlated risk and exposes liquidity providers to adverse selection when expanded to prediction market events. We built a 5-layer NLP and ML correlation detection pipeline that identifies shared entities, logical implications, and category relationships between markets, feeding into a Gaussian copula model that correctly prices joint probability. Every position is packaged as an EIP-712 signed certificate and verified against 18 independent on-chain risk checks before a single dollar moves. Liquidity providers deposit USDC into an ERC-4626 vault with a 7-day withdrawal delay for bank-run protection, earning yield from defaulted positions and protocol fees. Investors receive on-chain NFT receipts for their positions, with future features including a P2P secondary marketplace, thematic Composites, cross-margin matching, and a junior/senior tranche model all on the roadmap. Laytus is built on Robinhood Chain with a Python FastAPI microservices backend, React frontend, and a continuously hardened ML pricing model that ingests real-time adversarial flow.